Information Processing Bias in Post-traumatic Stress Disorder

4.1. Cognitive Psychology

4.2. Cognitive Electrophysiology

Event-related potentials (ERPs) are employed to measure brain responses to specific events, often auditory and visual stimuli, with a high temporal resolution (msec). The stimulus evoked responses, such as the P50 and P1/N1/P2 complex, indicate early stages of processing in sensory pathways and cortical areas (within 100 msec), while later endogenous ERPs, such as the N2/P3 complex, indicate more complex cortical processing. The activity indicated by these ERPs is modulated by cognitive processes such as attention and memory [80-93]. Cognitive ERP studies investigate the components of sensory processing and attention that are affected by trauma sensitivity in PTSD. These biological indices provide insight into the stages of information processing, from sensation to perception and memory, that are activated by traumatic information in PTSD [see also 94].

Attias, Bleich, and Gilat [95] investigated sensitivity to trauma in Israeli combat veterans using a visual oddball task with common pictures (furniture and flowers, 60%), rare target pictures (domestic animals, 20%) and rare combat pictures (20%). Combat veterans with PTSD had larger N1 and P3 amplitudes and shorter P3 latency for the combat pictures [see also 96, 97]. An increase in N1 amplitude may reflect an early, automatic allocation of attention to identification and discrimination of the physical attributes of stimulus information; in this case, it suggests that PTSD patients automatically orient to traumatic images. In addition, larger P3 amplitude for combat pictures indicates greater attention to evaluation of the significance or meaning of the traumatic images, which depends on prior exposure and sensitivity to their traumatic content. Also, P3 latency for traumatic images was positively associated with the frequency of traumatic intrusions. Furthermore, a discriminant function, based on P3 amplitude in response to combat pictures alone, correctly identified 85-90% of veterans with PTSD and 95% of veterans without PTSD. These results clearly indicate sensitivity to traumatic information in PTSD, at several levels of attentive processing.

However, these effects were not entirely specific to traumatic information. The P3 activity in PTSD was enhanced for both neutral targets and combat distracters; controls had larger P3 activity for neutral targets than combat images [96]. Also, there was some indication that traumatic images might interfere with processing neutral targets, because there was delayed P3 and reaction times for the neutral targets in PTSD [96]. These results may provide partial support for the theory that activation of a trauma schema in PTSD can enhance traumatic, but interfere with neutral information processing [58, 59].

Several studies have investigated the sensitivity of PTSD patients to both traumatic and non-traumatic emotional stimuli [98, 99] [see also 100]. Firstly, Kounios et al. [98] reported an ERP study of visual word stimuli that comprised common words (neutral or combat related) and rare target words (foods). Their findings indicate larger P3 amplitude for both combat and neutral words in PTSD (there was no difference in P3 activity between combat and neutral words). As in the previous work [96], this study also indicated larger P3 for all stimuli in the presence of traumatic stimuli. It is interesting that the frequency of combat stimuli was not a significant factor in eliciting enhanced responses; Attias et al. [96] presented combat stimuli rarely, whereas Kounios et al. [98] presented frequent combat stimuli. Also, there appear to be no substantial differences between pictorial and word stimuli. In contrast, Stanford et al. [99] observed greater P3 activity for trauma words and smaller P3 activity for neutral words in PTSD patients. They observed an enhanced frontal P3 for trauma words [it was specific to trauma, it did not generalize to social threat words; cf. 79]. Thus, while these studies to date confirm enhanced responses to trauma in PTSD, they are inconclusive with respect to associated effects on concurrent neutral information [cf. 58, 59].

However, even this conclusion is compromised by one ERP study of emotional Stroop words, which indicated smaller and later P3 activity for all words in PTSD patients, even for trauma words [101]. For all combat veterans, this study confirmed delayed reaction times for traumatic words, which has been observed in emotional Stroop studies [e.g. 71]. It also demonstrated that trauma words are related to greater frontal P3 than neutral words (for both PTSD and control subjects). This larger frontal P3 may confirm that delayed responses to emotional Stroop words reflect greater attention for the emotive words. However, the PTSD subjects had smaller and later parietal P3 ERPs than control subjects, so the PTSD patients were not evaluating those words as much as control subjects. This finding is inconsistent with the previous literature and theory on attention bias in PTSD, which clearly predicts greater responses for trauma words in PTSD. A large proportion of research findings appear to support this hypothesis, so it is difficult to reconcile this study with both theory and previous work. Perhaps the differences between studies reflect variations in the degree of intrusive or avoidance symptoms in patients or the nature of control groups. The differences might also depend critically on the absolute or relative frequency of presentation for neutral, emotive, threat, or trauma stimuli [c.f. 97]. Also, the nature of the stimulus information may be critical; perhaps PTSD patients are more sensitive to visual or auditory sounds associated with trauma than linguistic associations [50]. Further independent replication of this study would be valuable, with further consideration of intrusive and avoidance symptoms, control subjects, and frequency of stimulus presentations.

Kounios et al. [98] also found smaller P100 ERP amplitude over the right posterior temporal region in PTSD, for a series of visual words that contained traumatic and neutral words (the observed effects were not differentiated for traumatic and neutral words). Their results indicate that there are diminished early visual responses for all words in the presence of traumatic information. The findings of Kounios et al. [98] are inconsistent with previous reports of enhanced N1 activity for combat pictures [i.e., 95, 96]. Thus, their results also indicate there are decreased responses during early processing of visual word stimuli in PTSD, even if the stimuli are traumatic.

4.3. Cognitive Neuroimaging

Functional neuroimaging includes measures of regional cerebral blood flow (rCBF) from positron emission tomography (PET) and functional magnetic resonance imaging (fMRI). There are now several studies of rCBF during traumatic cognition in PTSD [114, 115, 117] [see also 3, 50, 118].

Rauch et al. [114] used PET to investigate rCBF in PTSD during imagination of either traumatic or neutral experiences, prompted by listening to a tape recording of their own narration. For this traumatic script-driven imagery, PTSD patients had increased rCBF in limbic and paralimbic structures of the right hemisphere, including medial orbitofrontal, anterior cingulate, insular, and anterior and medial temporal cortex, as well as the amygdala [114] [see also 119]. The orbitofrontal cortex and the anterior cingulate are engaged in the appraisal of emotional experience and monitoring conflicting demands [120, 121]. The amygdala, in particular, is involved in detection of threat and generation of fear, anger and associated autonomic and neuroendocrine responses [122, 123]. Another increase in rCBF was identified in secondary visual cortex, possibly related to traumatic visualization. In contrast, decreases in rCBF were found in left inferior frontal cortex (Broca’s area) and left middle temporal cortex, suggesting decreased linguistic cognitions during traumatic imagery [114, 115] [see also 50, 124, 125].

These seminal findings were explored further by Shin et al. [115], who investigated rCBF during (a) imagery of trauma, negative and neutral experiences and (b) visual presentations of trauma, negative and neutral stimuli (which were accompanied by an auditory description of the scene). This study did not identify increased activity in PTSD for combat pictures, compared with neutral pictures. Rather, there was decreased activity for PTSD patients in left middle frontal and anterior cingulate cortex. In contrast, the combat veterans without PTSD demonstrated increased activity for combat pictures in left middle frontal gyrus, Broca’s area, left superior temporal gyrus and left supramarginal gyrus (which may indicate greater auditory-verbal processing for the emotive combat pictures).

These findings of Shin et al. [115] are inconsistent with those to be expected, given theory on trauma sensitivity and prior findings of enhanced trauma responses in ERP studies [e.g., 95]. Perhaps the findings reflect avoidance in the PTSD patients. The PET procedure for blocked stimulus delivery is a continuous presentation of combat imagery, which may evoke an avoidance state in the PTSD subjects. For example, Bleich [97] observed that PTSD patients avoid processing repeated traumatic stimuli. They found that repeated combat pictures resulted in a rapid reduction in P3 amplitude and an increase in P3 latency for PTSD subjects. In most ERP tasks, traumatic images are rare and they occur among various image classes, which reduces or eliminates the effects of repetition suppression. Hence, a possible explanation for these results of Shin et al. [115] may be that PTSD subjects avoid the repetitive traumatic stimuli (perhaps a single-trial fMRI design could be adopted to evaluate this).

Another component of the work from Shin et al. [115] was a mental imagery condition, where patients voluntarily engage traumatic cognitions. In this study, there was increased anterior cingulate activity in PTSD for combat imagery, compared with neutral imagery; a confirmation of the previous findings [114]. Furthermore, comparison of responses to combat imagery versus combat pictures revealed enhanced activation in the left hemisphere, including the middle frontal gyrus, ventral anterior cingulate gyrus, and the supramarginal gyrus. There was also enhanced activation in the right amygdala. This suggests that previous findings of anterior cingulate and amygdala activity for traumatic imagery [114] could be specific to the voluntary generation of traumatic cognitions.

Further work has confirmed enhanced responses in the amygdala for fearful expressions. Rauch et al. [126] investigated fMRI of combat PTSD during a masked facial presentation, where facial expressions were either fearful or happy. These emotive faces were displayed very briefly (30 ms) and then masked by a longer neutral face expression (170 ms). This task design prevents controlled processing of the masked faces, thereby activating automatic emotional responses, with enhanced activity in the amygdala for the fearful faces. The results clearly demonstrated greater activation of the left amygdala for fearful faces, with PTSD patients demonstrating larger activity than combat veterans without PTSD [126] [see also 113].

4.4. Psychophysiology

The activation of a trauma schema is associated with a peripheral fear or stress response and the feedback from that response reinforces the impact of the initial cognitive and affective appraisal. Psychophysiology studies of PTSD indicate that traumatic cues and mental imagery are often related to heightened startle responses and peripheral hyperactivity, including increases in heart rate and blood pressure, muscle tension, and skin conductivity [127-139].

Several research groups have pursued the use of objective measures of physiological arousal, in response to traumatic stimuli, to discriminate war veterans with and without PTSD. In a series of studies, Blanchard and colleagues have successfully discriminated veterans with or without PTSD on the basis of heart rate responses to combat sounds [62% correct [128]; 95% correct for veterans with PTSD and non-veteran controls [127]; 75%-83% correct [134]; 81% correct [140]; 86.4% correct [129]; [see also 141, 142]]. They measured physiological responses to three conditions: (a) resting, (b) mental arithmetic, and (c) a combination of music, silence, and combat sounds that increase in loudness from 40 to 80 dB, in steps of 20 dB.

Blanchard et al. [134] found that a discriminant function, based on heart rate alone, correctly identified 84% of combat veterans with PTSD and 75% of all veterans. Furthermore, the same discriminant function correctly identified 83% of another sample of combat veterans with or without PTSD. Gerardi et al. [140] reported this discrimination has good short term test-retest reliability. Also, they reported that veterans with PTSD cannot consciously lower their physiological responses, so they appear to be without PTSD [cf. 139]. Also, although veterans without PTSD can elevate their responses to combat stimuli, it is still possible to discriminate them from veterans with PTSD with 85% accuracy using a discriminant function based on heart rate measures alone. Also, Casada [141] found that heart rate, more than electrodermal activity, was the most specific indicator of physiological arousal for traumatic content in PTSD.

In contrast, Pitman et al. [130] found that electrodermal activity was a better indicator of PTSD than heart rate. They assessed heart rate, forehead electromyogram (EMG), and skin conductance in response to unique combat incidents recollected by combat veterans. Independent raters agreed that the traumatic content and severity of the events recalled were similar for veterans with or without PTSD. The physiological response to the unique combat imagery was greater in veterans with PTSD than veterans without PTSD. In particular, skin conductivity (73%), rather than forehead EMG (67%) or heart rate (64%), provided the best predictive discrimination of veterans with or without PTSD. A step-wise discriminant function, based on all three measures, correctly classified all PTSD veterans. While the sensitivity was excellent, the specificity was not perfect; it had 79% accuracy for all veterans with or without PTSD. Also, physiological arousal was associated with the frequency of intrusions related to those traumatic events. Pit-man et al. [130] proposed that veterans with PTSD were “reliving” their traumatic experiences, whereas veterans without PTSD were merely “recollecting” their experience [cf. 139].

The discrepancy in the predictive value of skin conductance and heart rate between the research studies above might be partially explained by an increase in skin conductivity to specific traumatic imagery. Pitman et al. [130] found that skin conductance, but not heart rate, was greater in response to specific combat imagery than general combat imagery [see also 139]. There is contradictory evidence, however, in the results of Casada [141], which indicate that heart rate, more so than skin conductance, is a more specific indicator of traumatic arousal in PTSD [see also 139, 142].

Brende [143] reported that skin conductivity in response to traumatic imagery reflects the functioning of the contralateral hemisphere and that the left and right hemisphere play different roles in the symptoms of PTSD. Skin conductivity from the right hand is associated with hypervigilance and emotional numbing in PTSD and it reflects the function of the left hemisphere, which is related to logical, linguistic operations on information from the senses and the maintenance of high levels of vigilance for changes in stimulation [143]. The relationship between hypervigilance and emotional numbing is not very clear. It could be expected that hypervigilance will produce irritability and volatile emotions rather than numbness. However, Brende [143] asserts that hypervigilance in the left hemisphere could involve lateral inhibition of the emotional processing of the right hemisphere. Conversely, skin conductivity from the left hand reflects the activity of the right hemisphere, which processes or regulates traumatic imagery and its emotional impact in PTSD [143]. Thus, skin conductivity may not only indicate the degree of physiological arousal, but also the functional specificity of the left and right hemispheres during traumatic imagery.

The findings of Pitman et al. [130] and Brende [143] suggest that skin conductance biofeedback could be useful in the treatment of PTSD [see also 137]. This biofeedback could help PTSD patients to control their physiological arousal, to decrease their anxiogenic feedback. The ability to reduce anxious physiological responses to traumatic imagery could be an important aspect of the recovery process. For instance, Boudewyns and Hyer [144] used the physiological assessment technique of Pitman et al. [130] and found that therapeutic reductions in symptoms are associated with decreases in heart rate and especially skin conductance responses to traumatic imagery immediately after therapy. Boudewyns and Hyer [144] also reported that an exposure or flooding therapy reduces heart rate and skin conductance responses to traumatic imagery.

Murburg [145] points out interesting peripheral relationships to central processes. For instance, traumatic visualizations may induce increased peripheral epinephrine concentrations that increase memory consolidation by stimulating the amygdala, which is important in emotional stimulus evaluation and memory [see also 146, 147]. Also, “responding to a stressor . . . may itself leave behind molecular ‘memory traces’ that so alter involved neural pathways as to predispose them to be more readily activated in the future” [145, p. 354]. Reduction of peripheral hypersensitivity during psychotherapy promotes recovery from PTSD, which highlights the importance of visceral reinforcement for the maintenance of traumatic schemata [144] [cf. 32, 148].

4.5. Summary

These research studies have demonstrated sensitivity to traumatic information in PTSD, which is apparent in both neurocognitive and peripheral responses. Cognitive studies demonstrate traumatic sensitivity in attention and memory tasks, which is positively associated with intrusive symptoms. ERPs are enhanced for traumatic stimuli during both early sensory (N1) and later stimulus evaluation (P3) processes. ERPs also indicate that PTSD patients fail to discriminate relevant targets from irrelevant traumatic distracters. The ERP studies do not indicate sensitivity for emotional information in general, the sensitivity appears to be specific to trauma. Stimulus novelty is an important aspect of research design, several ERP studies indicate greater novelty responses in PTSD, even for non-traumatic stimuli. Some studies do not demonstrate sensitivity to traumatic stimuli, possibly due to extended traumatic presentations that evoke avoidance strategies; the sensitivity to trauma may be observed only for brief trauma stimuli or voluntary traumatic imagery. Neuroimaging during voluntary traumatic imagery demonstrates increased activity in limbic and paralimbic areas (the amygdala, medial orbitofrontal, anterior cingulate, insular, and anterior and medial temporal cortex). These areas are engaged in emotional evaluation, learning and memory. Peripheral stress feedback into these areas may serve to maintain and consolidate trauma sensitivity.

5.1. Hippocampal Volume in PTSD

Recent work demonstrates abnormal hippocampal anatomy in PTSD, which has important implications for cognitive functions in PTSD [145, 153, 156-166]. Functional neuroimaging of the hippocampus demonstrates its role in novel stimulus evaluation, episodic memory, and the spatio-temporal coherence of experience [145, 153, 156-166].

Several studies have now employed high-resolution magnetic resonance imaging (MRI) to measure the volume of the hippocampus in PTSD. The studies reported an average 10% reduction in hippocampal volume, with associated deficits in short-term memory [153, 156-158, 167, 168] [cf. 169]. These studies involve careful anatomical judgment during the selection and measurement of hippocampal regions, so it is important that they involve blinded research designs [167]. The hippocampal atrophy is smaller in PTSD than amnesia. For example, Press et al. [170] reported a 49% reduction of hippocampal volume in amnesic patients (with normal parahippocampal volumes). These patients showed severe deficits of verbal and non-verbal memory, while performing normally on other tasks [170]. Nevertheless, the functional significance of the hippocampal abnormality in PTSD has prompted careful examination of this structure.

Bremner et al. [153] found an 8% reduction of right hippocampal volume in Vietnam veterans with PTSD, which was associated with deficits in verbal short-term memory [see also 171]. Gurvits et al. [157] reported bilateral hippocampus reduction in combat PTSD, after controlling for age and whole brain volume. The decrease was greater in left (26%) than right (22%) hippocampus, especially after controlling for alcohol abuse and combat exposure. Hippocampal volume was negatively correlated with both combat exposure and PTSD symptom severity. Furthermore, hippocampal volume was positively correlated with several measures of attention and memory (digit span, arithmetic [WAIS-R], attention index [Wechsler memory scale], and delayed recall errors [Benton visual retention test]).

Several studies demonstrate smaller hippocampus in adults with a history of childhood abuse [156, 158]. Bremner et al. [156] found a 12% smaller left hippocampal volume in adult survivors of chronic childhood abuse (7 to 15 years of physical, sexual or emotional abuse), with 10% of this effect solely related to PTSD diagnosis (after sex, age, education or alcohol abuse were controlled). In a similar population of women who survived childhood sex abuse, Stein et al. [158] found a 5% reduction of the left hippocampus. Bremner et al. [156] found no correlations between left hippocampal volume and verbal memory impairments (immediate and delayed recall and retention), trauma onset or duration, years since trauma cessation, or PTSD symptom severity [cf. 157]. Bremner et al. [156] propose that trauma early in life may damage hippocampal structures, but neural plasticity could recover the integrity of verbal memory processes. In this regard, they propose that some functions of the hippocampus may have shifted to the left temporal lobe, which was larger in patients [156]. However, it is also possible that larger left temporal lobes could be related to better visual performance, which might be an adaptation to a loss of verbal capacity [156].

There are some concerns that decreases in hippocampal volume in chronic PTSD may result from substance abuse or other confounds [e.g., 169, 172], and it is not clear whether smaller hippocampal volumes precede or follow a traumatic experience [171, 173]. One longitudinal study has reported no decrease in hippocampal volume between 1 week and 6 months after trauma [167]. This study had tight controls over substance abuse and it employed a blinded measurement design, whereby the persons measuring hippocampal volume were blind to both the group and time status of the data. However, this study only investigated subjects older than 20 years, after which normal development of the hippocampus and other limbic structures is complete, whereas some previous studies have investigated patients who experienced trauma during childhood or early adulthood. Nevertheless, this study identified abnormal attention and executive functions, despite a lack of clear evidence for hippocampal atrophy, suggesting that executive networks are already dysfunctional in the first 6 months of PTSD [167]. The possible conclusions are: (a) there is no hippocampal atrophy and previous findings are confounded, or (b) there is atrophy, but it may take longer than 6 months to appear, so atrophy is only apparent in chronic patients, or (c) a smaller hippocampus might predispose some people to PTSD and especially chronic PTSD [173].

With regard to (c), another study has investigated whether smaller hippocampal volume may indicate a susceptibility to PTSD. Gilbertson et al. [168] compared combat veterans with and without chronic PTSD, together with their monozygotic twins who had no combat exposure. They found 10% smaller hippocampal volume in the PTSD twin pairs and there was a significant negative relationship between hippocampal volume and the severity of PTSD in the combat veterans with PTSD (r = –0.64) and their twin brothers (r = –0.70). The hippocampal volume effect remained significant after controlling for whole brain volume, age, combat exposure, and noncombat trauma incidents (including childhood physical or sexual abuse). There were no significant differences in total brain volume or amygdala volume. They conclude that smaller hippocampus could be a trait indicator of the likely development of PTSD after traumatic exposure [168]. However, smaller hippocampal volume was concentrated in the severe PTSD cases. The group differences in hippocampal volume only apply to severe PTSD after combat exposure (CAPS > 65, based on a subsample of 24 from 34 combat PTSD cases). Also, the combat veterans with PTSD had a history of more alcohol abuse than any other group in the study, so the results could be confounded by alcohol abuse. Thus, this study may provide evidence that combat exposure is not a necessary condition for hippocampal atrophy, but further longitudinal studies are required to confirm this conclusion [168] [see also 174].

The findings of abnormal neuroanatomy in PTSD may be specific to hippocampal volume, rather than a general or diffuse neurological abnormality. Bremner et al. [156] noted no significant differences in several regions, including caudate and amygdala. Similarly, Gurvits et al. [157] found no differences in intracranial cavity, whole brain, ventricles, ventricle:brain ratio, or amygdala (although, the right amygdala was larger in patients than controls, which approached significance at .07). Likewise, Gilbertson et al. [168] found no differences in whole brain or amygdala volume.

However, there are two reports of extra-hippocampal abnormality in PTSD. The first study indicated focal white matter lesions in only 8 of 42 combat related PTSD patients [175]. The lesions were in periventricular regions or near the white/gray cortical junctions. The lesions were not associated with symptom severity or comorbid depression or alcohol abuse. Rauch et al. [176] conducted a more comprehensive cortical parcellation in nurses with combat exposure. Although whole cortical volume was normal, this study clearly identified smaller cortical volume in the anterior cingulate and subcallosal cortex [176]. Some near significant decreases were identified in the angular gyrus, inferior occipito-temporal gyrus and the supplementary motor cortex [176].

In summary, evidence indicates reductions of hippocampal volume in chronic PTSD, where hippocampal volume is often positively associated with deficits in attention and short-term memory, combat exposure and the severity of symptoms. Some evidence suggests that chronic childhood abuse may contribute to hippocampal atrophy, but these observations are not associated with cognitive impairment, possibly due to neural plasticity and recovery of function. One longitudinal study of trauma failed to observe hippocampal changes after 6 months. Furthermore, this study observed deficits in cognitive functions, despite no evidence of hippocampal atrophy. Perhaps functions that involve the hippocampus are not necessarily related to the volume of the tissue, given plasticity and the distributed networks involved in complex cognition.

5.2. Hippocampal Physiology and Stress

The CA3 region of the hippocampus is a central feedback target for glucocorticoids, which regulate the hypothalamic-pituitaryadrenocortical (HPA) response to stress. Under stressful conditions, glucocorticoid levels increase, which is detected in the CA3 area of the hippocampus (among other locations of the HPA axis). Given this stimulus, the hippocampus can inhibit the hypothalamus from discharging corticotrophin releasing factor, thereby decreasing further cortisol responses to stress [145].

An explanation for abnormal hippocampal volume in PTSD is related to glucocorticoid physiology [e.g. 177]. Chronic, elevated levels of glucocorticoids can induce dendritic loss and cellular atrophy in the hippocampus [178-182] [see also 171, 183]. Glucocorticoids can interfere with the reuptake of glutamate from the synaptic cleft, resulting in excess stimulus of post-synaptic N-methyl-D-aspartate (NMDA) receptors. This can result in an excess of intracellular calcium, which has a variety of dangerous metabolic activities. So, an excess of glutamate and NMDA activation counteracts the normal process of calcium extraction and cellular repair. This physiology is related to several explanations for hippocampal abnormality in PTSD, including (a) hippocampal dendritic pruning, (b) cellular atrophy or (c) inhibition of neurogenesis [171, 182].

However, several studies of glucocorticoid levels immediately after exposure to trauma indicate that decreased cortisol predicts development of PTSD, whereas increased cortisol predicts onset of depression [8, 183]. This evidence is inconsistent with the hippocampal atrophy hypothesis of elevated cortisol levels in PTSD. There are two ways to reconcile these findings: (a) hippocampal changes do not arise immediately after trauma, but take at least several weeks [182], or (b) there is an explanation for hippocampal atrophy that does not depend on elevated cortisol levels. This explanation refers to evidence of greater glucocorticoid receptor density and sensitivity in the HPA feedback circuits among people who develop PTSD, suggesting an abnormal sensitivity to stress and more negative feedback on the HPA endocrine response [183].

It is important to understand hippocampal pharmacology and it’s relations with other cortical systems for cognition. If the hippocampal elements of the HPA feedback circuits are impaired, there should be less habituation of fear and stress responses. A possible pharmacological mechanism involves acetylcholine, which is involved in fear conditioning to context [184]. Acetylcholine is important for hippocampal theta rhythms, which are thought to synchronize distributed neural systems and facilitate the integration of sensory information into context [184]. Furthermore, there may be important interactions between acetylcholine and NMDA receptors, which are implicated in hippocampal dendritic pruning, atrophy and neurogenesis processes [171, 172, 181-184]. Also, the involvement of NMDA and serotonin have been demonstrated by inhibition of stress induced atrophy by application of tianeptine and phenytoin [181, 185, 186]. The effects on hippocampal dendrites related to glucocorticoids and NMDA physiology can be observed after 14 days, suggesting a long-term response process [182]. Acute stress can promote a short-term increase, but a long-term depression in acetylcholine activity through modulation of gene expression [29, 187]. The impact on cognition is also important, as cholinergic systems play a role in attention, learning and memory. For example, there is evidence of greater susceptibility to distraction and novelty with decreases in hippocampal cholinergic activity [159] [cf. 145]. Similarly, neuroimaging implicates abnormal noradrenergic activity in PTSD. One PET study of an adrenergic stimulant (yohimbine) showed abnormal hippocampal activity, as well as prefrontal, orbitofrontal, parietal and temporal cortex [188]. Perhaps these cellular processes provide an explanation for the impact of trauma on contextual stimulus evaluation and learning in PTSD, which consists of generalization from one context to many contexts and failure to habituate. The physiological disturbances need not be associated with abnormal hippocampal volume [e.g. 172]; even normal levels of glucocorticoids can induce dendritic remodeling [182] [cf. 183]. The effects of stress may be related to neurotransmission adaptations and gene expression [29, 187, 189, 190] [see also 32, 148]. Nevertheless, it seems to be a disturbance of hippocampal functions that play an important role in the response to trauma in PTSD.

6.1. Cognitive Neuropsychology

As discussed above, neuropsychological theory proposes that abnormal limbic and cortical interactions contribute to difficulties with perceptual discrimination and integration of new information into episodic memory [66, 151, 152, 154] [see also 4]. Studies of PTSD demonstrate deficits of attention and executive functions [191-200] [see also 169, 193, 201, 202]. These deficits are observed in the performance of various tasks, including a continuous performance task, verbal fluency, comprehension, similarity judgments, digit-symbol substitution, and automatic tasks such as counting backwards, counting forwards by threes, and reciting the alphabet. Several studies also demonstrate memory impairments in PTSD; including immediate memory impairments in both verbal and visual tasks [191, 192, 203, 204] and some studies identify long-term memory deficits for verbal tasks and famous events [191, 204], but results for visual tasks are equivocal [192, 200, 204, 205]. Also, several studies have identified deficits of auditory verbal learning and memory [26, 192] [cf. 201]. Gurvits et al. [195] indicate that cognitive impairment may precede the onset of PTSD [cf. 201]. Gilbertson et al. [198] compare various neuropsychological measures and they argue that abnormal memory for neutral information is a critical determinant of PTSD diagnosis. Thus, neuropsychology studies employ trauma-neutral tasks and they most often demonstrate impairment of diverse executive cognitions in PTSD.

6.2. Cognitive Electrophysiology

Cognitive electrophysiology during neutral information processing provides insights into the sensory discrimination, attention and working memory processes of PTSD [see also 94]. The electrophysiological studies to date have largely focused on auditory discrimination.

Paige et al. [133] employed ERPs to investigate responses to increasing auditory intensity in PTSD. Veterans with PTSD show normal increases in N1, but abnormal decreases in P2 amplitude as auditory intensity increases toward and above the startle threshold (the effect was more prominent in the left than the right temporal area). Paige et al. [133] propose that a P2 reduction indicates “protective inhibition” -PTSD patients inhibit stimulation that could overload their capacity for accurate discrimination. Lewine [206] replicated this study with combined EEG and magneto-encephalography (MEG); they found P2 reductions in the auditory association cortex and confirm that the effect is a decrease in cortical activity rather than variation in the location of cortical source activity. These findings suggest a cortical abnormality of auditory processing in PTSD.

Recent studies suggest abnormal sensory habituation in PTSD, indicated by early auditory P50 ERPs [207-209] [see also 138]. Neylan et al. [208] reported a sophisticated investigation of P50 habituation for non-startle tones and found less habituation in PTSD patients [see also 207]. Neylan et al. [208] propose that the reduced P50 habituation could be related to deficits in hippocampal functions, which normally inhibit the response of thalamic circuits to repetitive, innocuous stimuli [many studies have implicated the hippocampus and associated limbic structures in PTSD, [145, 153, 156-166]. These studies show that PTSD patients do not habituate the P50 response for non-startle repeated clicks, suggesting an abnormality of sensory gating in brainstem and thalamic circuits.

This evidence can be interpreted in the context of hypotheses about abnormal interactions among subcortical, limbic and cortical networks [66, 151, 152, 154]. In particular, the evidence supports the proposal that trauma leads to hyperarousal or hypersensitivity of sensory systems. If brainstem and thalamic circuits do not habituate, their cortical projections are overloaded by repetitive information. Kolb [66] proposed that excessive emotional arousal, especially threat and fear responses of the amygdala, may overload executive systems (e.g., anterior cingulate), leading to less control of subcortical and primary sensory cortical activity, with consequent deficits in cortical discrimination and adaptive response processes [210]. Primarily, thalamic nuclei become hypersensitive to any stimulus change, so their projections to the cortex can overload the capacity to process sensory information and impair corticothalamic feedback required for effective early discrimination and habituation. Furthermore, a vicious cycle of hyperarousal ensues, as the capacity of cortical processes to regulate brainstem arousal nuclei diminishes. Hence, brainstem arousal centers, such as the reticular system and the locus coeruleus, may escape from cortical control and further enhance the sensitivity of thalamic circuits and their associations with limbic and cortical networks. Thus, there could be abnormal interactions of thalamic circuits and cortical systems in PTSD, resulting in impaired neutral stimulus information processing.

This evidence complements the findings that patients have a lower threshold at which sensory discrimination is overloaded by intense stimulation, promoting cortical inhibition of excessive stimulation. The primary auditory cortex demonstrates a stimulus driven response, with N1 augmentation to increasing stimulus intensity, but the auditory association cortex demonstrates active inhibition of excessive stimulation [i.e., P2 reductions; 133, 206]. It may be only at the level of associative cortical control that the early gating deficits are counteracted [133]. Kolb [66] hypothesized that frontal cortical systems for executive control of subcortical brainstem and other thalamic circuits may be deficient. The efficient modulation of elementary sensory discrimination processes is important for adaptive stimulus information processing, so these initial insights into Kolb’s [66] hypothesis form an important contribution to understanding PTSD.

An interesting study of sensory processing in the absence of attention has some important implications for any theory of early sensory processing in PTSD. Morgan and Grillon [211] investigated sensory discrimination, using an auditory mismatch negativity (MMN) task that does not require active attention, for PTSD patients who suffered sexual assault. They found no indications of abnormal P50, N1 or P2 ERPs, but enhanced N2 amplitude, due to a larger MMN, in PTSD patients. The habituation of P50 activity is difficult to investigate and the report by Neylan et al. [208] employed sophisticated analysis procedures to clarify this issue, so their results may carry greater weight than other studies that are not specifically designed to evaluate the P50. Nevertheless, based on their evidence, Morgan and Grillon [211] conclude that the P50, N1 and P2 ERPs indicate no abnormality of auditory pathways and cortical processing in PTSD [which is consistent with studies of similar auditory discrimination tasks; see discussion below on 105, 212-216]. However, Morgan and Grillon [211] find that the enhanced MMN suggests greater activity related to auditory discrimination in PTSD [cf., 212, 215]. The MMN is thought to indicate the automatic detection of auditory stimulus changes, involving stimulus registration and comparison in sensory memory [e.g., 82, part 3]. This result suggests that previous findings of abnormal sensory discrimination in PTSD may be related to controlled attention modulations of sensory discrimination. That is, there may be abnormalities in the systems that control top-down modulation of sensory discrimination in PTSD.

These findings of Morgan and Grillon [211] should be replicated, due to concerns about comorbid panic disorder in that study. Nearly half of the PTSD sample had comorbid panic disorder (PD) with agoraphobia. It is possible that their findings of abnormal early sensory discrimination could be related to comorbid PD [217, 218]. Most studies of PD indicate abnormal activity of later, evaluation processing [219-224]. In one PTSD study, there is evidence of larger P3 activity for PTSD patients with comorbid PD [105]. It is not yet clear if this comorbidity has implications for early ERP activity, such as the MMN.

If there is abnormal early sensory processing in PTSD, it is not clear whether cortical control processes can recover the integrity of information processing. Several ERP studies demonstrate abnormal activity during attention and working memory tasks in PTSD. In an ERP study of PTSD, McFarlane et al. [212] employed a three-tone discrimination task with common tones (1000 Hz) and both rare distracter (500 Hz) and target tones (2000 Hz, all stimuli had a constant duration and sub-startle intensity). For these neutral tones, PTSD patients had a slower response time and both later N2 and smaller P3 ERPs for infrequent target tones, there was also no difference of P3 amplitude for target and distracter tones. These findings were replicated several times [213, 215, 225]. The findings implicate cortical processing engaged in sensory discrimination, stimulus evaluation and working memory processes. The increase in N2 latency suggests difficulty with auditory discrimination [226] [cf. 211]. The smaller P3 amplitude could reflect deficits in the evaluation of stimulus relevance or meaning [227, 228]. In a further study of the P3 evaluation process, subjects were required to detect consecutive repeats in a five tone discrimination task, which required frequent updating of the physical attributes of target tones [214]. Smaller P3 activity for common and target tones indicated deficits in working memory processing for PTSD patients [214, 215]. This deficit was further investigated in the visual modality, which also indicated abnormal frontal and parietal ERP topography during working memory in PTSD [216, 229]. These indications of abnormal attention and working memory processes can be related to the capacity to regulate traumatic intrusions in PTSD [48]. Thus, these studies extend the findings of abnormal early sensory processing to later aspects of cortical processing engaged in stimulus discrimination, evaluation and working memory integration.

The decrease of P3 amplitude may be related to catecholamine dysfunction in PTSD. Kolb [66] hypothesized that sensory discrimination is affected by catecholamines, providing an important theoretical link between neurophysiology and cognition [see also 10, 154]. There is evidence that catecholamine neurotransmission modulates the capacity for attention [230-234]. An indication that catecholamines can normalize deficits of P3 activity in PTSD was found in studies of PTSD related to combat and sexual abuse [105, 235]. These findings indicate that catecholamine medications may normalize brain activity related to attention processes in PTSD and they provide partial support for Kolb’s [66] hypothesis of catecholamine effects on cognitive processes in PTSD.

These findings of abnormal P3 in PTSD can be modified by symptom severity and comorbidity. There are decreases in P3 and slower reaction times with dissociative and numbing symptoms [103, 104], as well as depression symptoms [105]. There are increases in P3 with panic symptoms [105, 219, 235]. Furthermore, the abnormality in P3 processes is not specific to PTSD [236], it has been observed in diverse psychiatric conditions, which suggests a central role for these processes in integrated cognition.

The neutral stimulus processing in PTSD could vary with affective states. That is, attention bias theory would predict that during stressful or traumatic affect, the attention available for neutral stimuli decreases. However, Kaufman [103] found no effects of emotional state on P3 responses to neutral stimuli. Previous work has either included emotive or trauma content into task materials or ensured that all task materials and the emotional context are neutral [102]. Kaufman [103] employed a neutral auditory discrimination task with prior induction of an affective context. Across three affective contexts, neutral, stressful and traumatic, there were no modulations of the P3 effects for neutral auditory processing. However, the emotional context did have an impact on dissociative states, which were related to P3 effects for distracters [102, 103]. This study suggests that attention biases are not simply a result of general affective states, but rather they arise as specific responses to stimulus information.

The distraction in PTSD could be dependent on the novelty of the information, because several studies demonstrate larger P3 activity for novel stimuli in PTSD. Blomhoff et al. [102] found larger frontal activity at 200-300 msec for rare, novel acoustic non-words in PTSD, which is very similar to P3a effects that indicate an orientation of attention to novel stimuli. They also found that this frontal activity between 150- 300 msec was positively associated with the arousal, avoidance and intrusion symptoms of PTSD. Similarly, Kimble et al. [237] investigated novelty sensitivity in combat veterans with or without PTSD. They used two auditory tasks. One was a three-tone oddball task, with 1000Hz common tones, 500Hz rare distracters and 2000Hz rare target tones. The second task was almost identical, except the 500Hz rare distracters were replaced with various novel sounds (clicks, buzzes, etc.). If PTSD patients are susceptible to novel distraction, they expected to see larger ERP activity for the novel stimuli. Also, they expected to see greater impairment of target processing in the novel task. In fact, their results show that all subjects did have smaller P3 activity for targets during the novelty task, indicating that novel stimuli diverted attention from the primary task. However, there were no group differences in this deficit for target processing2. There was a group by task interaction for distracters, which comprised greater frontal P3 amplitude for novel than pure tone distracters in PTSD. The effect identified was significant after adjustment for combat exposure and comorbid panic disorder (there was no correction for depression). This finding required comparisons between neutral and novel, ambiguous distracters. With similar task designs, it is possible to explore the effects of emotional distracters [c.f., 102, 238]. The novel sounds in these studies were not related to traumatic experience and therefore suggest that threat sensitivity in PTSD extends beyond traumatic stimuli to any ambiguous novel stimuli.

6.3. Cognitive Neuroimaging

Semple et al. [239, 240] found abnormal activity in parietal attention systems during neutral information processing in PTSD [see also 241]. They investigated rCBF in combat PTSD during an auditory continuous performance task (CPT), which required vigilant attention and discrimination for rare target events. PTSD patients had poor discrimination of targets and decreased rCBF in the angular gyrus of the right parietal cortex, a brain region that is implicated in multimodal attention processes [239, 240, 242, 243]. The parietal rCBF was positively associated with performance accuracy [239, 240]. Although the patient group had a comorbid substance abuse, the authors suggest their results are due to PTSD, because patients abstained for several weeks before the study and the PTSD results were inconsistent with results from another substance abuse group [239, 240].

Several recent papers demonstrate abnormal functional activity in PTSD for neutral stimuli [244-246]. Clark et al. [245] employed PET during two target detection tasks, using the same visual stimuli, with two different task instructions. In one task, a target word remained constant, while the other task required detection of repeated words (a variable target). The latter task required greater verbal working memory updating, as each non-target word became a new target template. In controls, activation was identified in a working memory network, including the dorsolateral prefrontal cortex, the orbitofrontal cortex and anterior cingulate, as well as the supramarginal gyrus of the inferior parietal lobe [246]. PTSD patients demonstrated less prefrontal activity, especially in the left dorsolateral prefrontal cortex, with possible compensation apparent in greater activity in the right parietal cortex [229, 244, 245]. This pattern of activity was taken to indicate deficits in prefrontal control of working memory content, which may be stored as multimodal representations in posterior parietal regions. The deficit in left prefrontal activity may be associated with decreased verbal skills in PTSD, which has been observed in studies of traumatic information processing [114]. In contrast with those studies of traumatic imagery, activation was identified in the middle frontal gyrus, which is implicated in speech production and more complex integration of information in working memory [247-250]. Thus, this work demonstrates a more fundamental difficulty with working memory for neutral information in PTSD, which is consistent with theories about the dissociation of traumatic memory from linguistic, narrative encoding [50]. Also, the integrity of working memory processes can be related to the capacity to control intrusive memories [48].

6.4. Psychophysiology

PTSD patients have high cardiovascular activity, including increased heart rates and diastolic and systolic blood pressure, during resting periods of laboratory research [128, 140], and before a routine outpatients physical examination [251]. Casada [141] found that PTSD patients have higher baseline electrodermal activity, but not heart rate, in comparison with combat and non-combat controls. Furthermore, PTSD patients have often demonstrated a heightened startle response to sudden loud tones, which is indicated by increased eye blink amplitude, heart rate, skin conductivity, and diminished habituation of skin conductivity over several tone presentations [132, 133, 135, 136]. An exception to these findings is that PTSD patients show no significant startle response to 2 sec of white noise presented at 80 dB [129]. PTSD patients may be less responsive to the white noise, which had a longer duration and a different sound texture, than the sudden tones presented in other studies.

The physiological hyperactivity in PTSD, while among neutral circumstances, cannot be explained as a conditioned response. Shalev et al. [135, 136] proposed that hyperarousal symptoms indicate a genetic disposition or an acquired sensitivity to autonomic reactivity and a diminished habituation of autonomic responses. Gerardi [251] proposed that PTSD patients may activate a cognitive fear network during any laboratory or hospital visit and this promotes physiological arousal.

6.5. Summary

Psychophysiology studies indicate hyperarousal in PTSD, even among relatively neutral circumstances. At the cognitive level, neuropsychology studies indicate deficits in various attention and memory tasks. Several ERP studies demonstrate abnormal thalamic gating and early sensory processing for auditory tones in PTSD. There is also evidence for abnormal stimulus discrimination, evaluation and working memory integration, indicated by abnormal N2 and P3 ERPs in PTSD. Finally, several studies of rCBF demonstrate abnormal activity in frontal and parietal systems engaged in attention and working memory processes.